g58892881

6 ms·

1.
Show HN: In-browser 4x photo upscaling(crisp.photos)
2 points by g58892881 1mo ago|0 comments
2.
by g58892881 1mo ago
Thank you
3.
by g58892881 1mo ago
Photo upscaling and inpainting apps, both running in the browser https://crisp.photos https://wipe.photos
4.
by g58892881 1mo ago
pi
5.
Show HN: In-browser photo inpainting with LaMa and ONNX(wipe.photos)
1 points by g58892881 2mo ago|0 comments
6.
by g58892881 2mo ago
also, i double checked, nano is nano and micro is micro. didnt fuck that up
7.
by g58892881 2mo ago
unable to fix it. latest dev ort, dispoing tensors, restarting the session/worker. nothing, crashes every time. defaulting to wasm on ios devices now
8.
by g58892881 2mo ago
right. happens on my 13 too. memory leak confirmed, not sure yet what's causing it
9.
by g58892881 2mo ago
wrote a web-demo, runs in your browser https://inflect-tts.geronimo-labs.com code: https://github.com/geronimi73/inflect-tts
10.
by g58892881 2mo ago
afaik controls random noise injected into the model's latent vector before decoding same seed + same variation + same text -> bit-identical audio
11.
Show HN: Inflect TTS v2+ONNX, 9M/4M text-to-speech models running in the browser(inflect-tts.geronimo-labs.com)
7 points by g58892881 2mo ago|2 comments
12.
by g58892881 3mo ago
well done! unet weights are in fp32. did you by any chance try something lower, fp16?
13.
by g58892881 3mo ago
An all time classic, a headshot generator: https://instant.photos
14.
by g58892881 3mo ago
https://crisp.photos In-browser 4x image upscaling. Vite + Onnxruntime + https://huggingface.co/Kim2091/UltraSharpV2
15.
by g58892881 4mo ago
Surprise
16.
FLUX.2 Klein – How Inference Works(medium.com)
2 points by g58892881 5mo ago|0 comments
17.
by g58892881 6mo ago
pretty basic ``` from torchvision import models # Avoid downloading pretrained weights; we load trained checkpoint weights. model = models.resnet18(weights=None) model.fc = nn.Linear(model.fc.in_features, num_classes) ``` https:/&#x2F
18.
by g58892881 6mo ago
I created a demo running in the browser, on your device: https://next-voice.vercel.app
19.
by g58892881 1y ago
my guess it will always be "some" some photos made by AI some books written by AI some code written by AI not all